CC(t;3) PSI4 Plugin

.. image:: https://travis-ci.org/piecuch-group/cct3.svg?branch=dev :target: https://travis-ci.org/piecuch-group/cct3

The CCT3 plugin to PSI4 [1]_ is capable of executing a number of closed- and open-shell coupled-cluster (CC) calculations with up to triply excited (T3) clusters. Among them is the active-space CC approach abbreviated as CCSDt [2], [3], [4], [5], which approximates full CCSDT by selecting the dominant T\ :sub:3 amplitudes via active orbitals, and the CC(t;3) method, which corrects the CCSDt energies for the remaining, predominantly dynamical, triple excitations that have not been captured by CCSDt [6], [7]. The CC(t;3) approach belongs to a larger family of methods that rely on the generalized form of biorthogonal moment expansions defining the CC(P;Q) formalism [6], [7].

The CCSDt method alone is already very useful, since it can reproduce electronic energies of the near-CCSDT quality at a small fraction of the computational cost, while accurately describing selected multireference situations, such as single bond breaking. CC(t;3) improves the CCSDt energetics even further, being practically as accurate as full CCSDT for both relative and total electronic energies, at a cost which is essentially the same at that of CCSDt. The systematic convergence of the CCSDt and CC(t;3) calculations toward CCSDT should be emphasized here too. For example, CCSDt becomes full CCSDT when all orbitals used to select T3 amplitudes are active. The same applies to CC(t;3) (when all orbitals used to select T3 amplitudes are active, the triples correction to CCSDt becomes zero).

The CCT3 plugin can also be used to run CCSD and CR-CC(2,3) calculations. This can be done by making the active orbital set, which is defined by the user in the input, empty, since in this case CCSDt = CCSD and CC(t;3) = CR-CC(2,3). We recall that CR-CC(2,3) is a completely renormalized triples correction to CCSD, which improves the results obtained with the conventional CCSD(T) approach without resorting to any multireference concepts and being at most twice as expensive as CCSD(T) [8], [9], [10]_.

Please note that UHF/UKS references are not supported by CCT3.

Build

To compile this plugin, a working version of PSI4 version 1.1 or greater is required. The easiest way to get a working copy is via the conda <https://conda.io/docs/>_ or anaconda <https://www.continuum.io/downloads#linux>_ environment (more on this here <http://www.psicode.org/psi4manual/1.3.2/conda.html#faq-psi4pkg>_). To use it, first create a new python 3 environment and activate it via:

.. code-block:: bash

$ conda create -n p4env python=3.7 psi4 psi4-dev -c psi4 -c psi4/label/dev $ source activate p4env

Next, get the source code for the CC(t;3) plugin and compile it.

Compiling using the highest optimization level (-O3) is recommended for best performance, but the compiler may use large amounts of memory (16+ GiB) during the build. If that is a problem, lower optimization levels may be used, but the speed of the resulting CCT3 binary may suffer.

If the machine has at least 32 GiB of RAM, you should have no problem executing the following lines:

.. code-block:: bash

$ git clone https://github.com/piecuch-group/cct3 $ cd cct3 $ psi4 --plugin-compile -DCMAKE_Fortran_FLAGS="-O3" $ make $ make install

Once this step is done, you should have a working copy of the plugin. You can run a test example with:

.. code-block:: bash

$ psi4 examples/H8-0.1.dat

Note: if you run into trouble importing the cct3 module, try adding the path of the directory containing the cct3 folder to your PYTHONPATH.

.. code-block:: bash

$ export PYTHONPATH="/path/to/cct3/parent/folder:$PYTHONPATH"

Run

In order to run a CCSD, CR-CC(2,3), CCSDt, or CC(t;3) calculation, the following options have to be set within the scheme

::

set psi4-cct3 { option value ... }

froz Number of frozen core molecular orbitals. act_occ Number of active occupied molecular orbitals counting from the Fermi level down (e.g. HOMO, HOMO-1, HOMO-2, etc.). act_unocc Number of active unnocupied molecular orbitals counting from the Fermi level up (e.g. LUMO, LUMO+1, LUMO+2, etc.). etol Energy convergence tolerance given as 10^-ETOL. Default is 10^-7 max_iter Maximum number of iterations. Default is 100. keep_amps If true, write down the converged cluster amplitudes to the file amplitudes.moe. calc_type Can be set to CCSD, CR-CC, CCSD3A, or CCT3. These options invoke CCSD, CR-CC(2,3), CCSDt, and CC(t;3) calculations, respectively. It not specified, the default is CCSD.

References

.. [1] \J.E. Deustua, J. Shen, P. Piecuch, "CCT3: A PSI4 Plugin Which Performs Active-Space Coupled-Cluster CCSDt Calculations and Which Can Determine Noniterative Corrections to CCSDt Defining the CC(t;3) Approach." .. [2] \P. Piecuch, "Active-Space Coupled-Cluster Methods," Mol. Phys. 108, 2987-3015 (2010). DOI: http://dx.doi.org/10.1080/00268976.2010.522608.

.. [3] \N. Oliphant and L. Adamowicz, "The Implementation of the Multireference Coupled-Cluster Method Based on the Single-Reference Formalism," J. Chem. Phys. 96, 3739-3744 (1992). https://doi.org/10.1063/1.461878. .. [4] \P. Piecuch, N. Oliphant, and L. Adamowicz, "A State-Selective Multi-Reference Coupled-Cluster Theory Employing the Single-Reference Formalism," J. Chem. Phys. 99, 1875-1900 (1993). DOI: http://dx.doi.org/10.1063/1.466179. .. [5] \P. Piecuch, S.A. Kucharski, and R.J. Barlett, "Coupled-Cluster Methods with Internal and Semi-Internal Triply and Quadruply Excited Clusters: CCSDt and CCSDtq Approaches," J. Chem. Phys. 110, 6103-6122 (1999). DOI: http://dx.doi.org/10.1063/1.478517. .. [6] \J. Shen and P. Piecuch, "Biorthogonal Moment Expansions in Coupled-Cluster Theory: Review of Key Concepts and Merging the Renormalized and Active-Space Coupled-Cluster Methods," Chem. Phys. 401, 180-202 (2012). DOI: http://dx.doi.org/10.1016/j.chemphys.2011.11.033. .. [7] \J. Shen and P. Piecuch, "Combining Active-Space Coupled-Cluster Methods with Moment Energy Corrections via the CC(P;Q) Methodology, with Benchmark Calculations for Biradical Transition States," J. Chem. Phys. 136, 144104-1 - 144104-16 (2012). DOI: http://dx.doi.org/10.1063/1.3700802. .. [8] \P. Piecuch and M. Wloch, "Renormalized Coupled-Cluster Methods Exploiting Left Eigenstates of the Similarity-Transformed Hamiltonian," J. Chem. Phys. 123, 224105-1 - 224105-10 (2005). DOI: http://dx.doi.org/10.1063/1.2137318. .. [9] \P. Piecuch, M. Wloch, J.R. Gour, and A. Kinal, "Single-Reference, Size-Extensive, Non-Iterative Coupled-cluster Approaches to Bond Breaking and Biradicals," Chem. Phys. Lett. 418, 467-474 (2006). DOI: http://dx.doi.org/10.1016/j.cplett.2005.10.116. .. [10] \M. Wloch, J.R. Gour, and P. Piecuch, "Extension of the Renormalized Coupled-Cluster Methods Exploiting Left Eigenstates of the Similarity-Transformed Hamiltonian to Open-Shell Systems: A Benchmark Study," J. Phys. Chem. A 111, 11359-11382 (2007). DOI: http://dx.doi.org/10.1021/jp072535l.

Development Team

| Dr. J. Emiliano Deustua | COO and Co-founder, Examol | e-mail: edeustua@gmail.com | | Dr. Jun Shen | Senior Research Associate, Department of Chemistry, Michigan State University | e-mail: shenjun@msu.edu | | Professor Piotr Piecuch | University Distinguished Professor and Michigan State University Foundation Professor, Department of Chemistry, Michigan State University
| Adjunct Professor, Department of Physics and Astronomy, Michigan State University